Document YDnz127j76Ep4n5V40vzJvGN

ELSEVIER Chemosphere 226 (2019) 898-906 Contents lists available at ScienceDirect Chemosphere journal homepage: www.elsevier.com/locate/chemosphere Waste incineration of Polytetrafluoroethylene (PTFE) to evaluate potential formation of per- and Poly-Fluorinated Alkyl Substances Chec updates (PFAS) in flue gas Krasimir Aleksandrov a, Hans-Joachim Gehrmann a, Manuela Hauser a, Hartmut Matzing a, Daniel Pigeon b. , Dieter Stapf a, Manuela Wexler a a Karlsruhe Institute of Technology, Karlsruhe, 76021, Germany b W.L. Gore & Associates, Inc., Elkton, MD, 21901, USA HIGHLIGHTS Municipal incineration of PTFE shows no significant generation of studied PFAS. Using pilot scale equipment and paired t-testing minimizes background interference. PTFE produced mainly hydrofluoric acid and carbon dioxide during incineration. ARTICLE INFO Article history: Received 7 January 2019 Received in revised form 28 March 2019 Accepted 31 March 2019 Available online 4 April 2019 Handling Editor: J. de Boer Keywords: PFAS Polytetrafluoroethylene Waste incineration Flue gas sampling Paired t-test PFOA/PFOS ABSTRACT In recent years, concerns over some per- and polyfluorinated alkyl substances (PFAS) have grown steadily. PFAS are a large group of chemical substances with widely differing properties. While one class of PFAS, fluoropolymers, have been demonstrated to meet the OECD criteria for polymers of low concern during the in use phase of their lifecycle, questions remain regarding waste handling at the end of useful life for products containing fluoropolymers. To show that polytetrafluoroethylene (PTFE) can be almost fully transformed into fluorine (F) (as hydrofluoric acid (HF)) and to study the possible generation of low molecular weight per- and polyfluorinated alkyl substances (PFAS), PTFE combustion under typical waste incineration conditions at the BRENDA (German acronym for "Brennkammer mit Dampfkessel") pilot plant at Karlsruhe Institute of Technology (KIT) was investigated. Results indicate that, within procedural quantitation limits, no statistically significant evidence was found that the PFAS studied were created during the incineration of PTFE. Therefore, municipal incineration of PTFE using best available technologies (BAT) is not a significant source of the studied PFAS and should be considered an acceptable form of waste treatment. 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.orgilicensesiby-nc-nd/4.0/). 1. Introduction Concerns over certain per- and polyfluoroalkyl substances (PFAS) (also called per-and polyfluoroalkyl compounds (PFCs)) have grown since the May 16th, 2000 USEPA press release announcement of the phase out of perfluorooctane sulfonate (PFOS) due to its toxicity, environmental persistence and bioaccumulation (USEPA, 2000). This class of compounds have been found throughout the environment from a variety of industry and consumer sources * Corresponding author. E-mail address: @wlgore.com (D. Pigeon). (Prevedouros et al., 2006; Rankin et al., 2016; Taniyasu et al., 2005). Today, many PFAS are under scrutiny, including perfluoroalkyl carboxylic acids (PFCAs), perfluoroalkyl sulfonates (PFSA5), perfluorooctane sulfonamides (PFOSAs) and perfluorooctane sulfonamidoethanols (PFOSEs). In addition to the manufacturing and intentional use of these compounds, other potential pathways where these PFAS might be generated are being studied (Prevedouros et al., 2006). One potential pathway identified for investigation is the waste handling of fluoropolymers at the end of useful life, specifically the municipal incineration of PTFE. Due to its unique properties, PTFE is used in a wide range of products including wire insulation, gasket material, filtration and waterproof garments (Henry et al., 2018). At the end of useful life, https://doi.org/10.1016/j.chemosphere.2019.03.191 0045-6535/ 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.orgilicensesiby-nc-nd/4.0/). K. Aleksandrov et al. / Chemosphere 226 (2019) 898e906 899 these products are subject to different waste streams including landfilling which accounts for 56% w/w of waste treatment worldwide and incineration (IEA, 2014). The number of waste incineration facilities are increasing globally as the best available waste treatment technologies can be applied (Gehrmann et al., 2017; EC, 2010; 17 BImSchV, 2003). Although PTFE is inert in the environment due to its high chemical and thermal stability, municipal waste incinerators generate adequate temperatures to decompose PTFE (Taylor et al., 2014). This study investigates the possible generation of a wide range of PFAS (Table 1) from PTFE incineration under standard municipal waste conditions. Extensive investigations at BRENDA pilot plant at KIT were conducted to validate that PTFE can be almost fully transformed to fluorine as hydrofluoric acid (HF) and a number of trace species in very low concentrations via incineration using the BAT. 2. Materials and methods Due to the environmentally ubiquitous nature of the substances listed in Table 1 and the extreme sensitivity of the liquid chromatography with mass spectrometry (LC-MS/MS) detection methods, contamination of solvents, samples and blanks was a significant concern (Prevedouros et al., 2006; Sinclair et al., 2007; Taniyasu et al., 2005). To reduce the probability of producing false positive results, a three parallel step approach was taken. First, the experiment was scaled up to pilot plant incineration levels using the BRENDA facility (Fig. 1). The solid combustion material input was many orders of magnitudes larger than in lab based incinerator simulations. Second, to minimize potential external contaminates, combustion input materials were limited to natural gas, commercial premium wood pellets, PTFE polymer pellets and air. In addition, paired t-testing was used to identify the presences of statistical differences between blank and PTFE spiked conditions (Van Belle et al., 1993). For this study, compounds were chosen to represent a broad range of PFAS. Specific compounds in Table 1 were selected due to their occurrence in the environment, literature citations and availability of validated methods from commercial laboratories (Prevedouros et al., 2006; Sinclair et al., 2007; Rankin et al., 2016). While some of the compounds listed are less likely to form from the incineration of PTFE, the perfluoro-carboxylic acids, thirteen of which were included in this study, have been suggested as potential combustion products (Arito and Soda, 1977; Ellis et al., 2003). 2.1. BRENDA facility The Institute for Technical Chemistry at KIT operates a rotary kiln test facility equipped with a boiler for heat recovery and a flue gas cleaning system which complies with German emission regulations (17 BImSchV, 2003). The pilot plant BRENDA (Fig. 1) provides scalable combustion research opportunities such as thermal behavior of end-of-life technical and consumer products. BRENDA has an overall thermal power of 2.5 MW, where 1.5 MW are from the rotary kiln and 1 MW from the post combustion chamber. (Nolte et al., 2005). For this study, PTFE and wood pellets were weighed and fed to the rotary kiln, while natural gas was supplied to the kiln and to the post combustion chamber. Table 1 in the Appendix summarizes all experimental process parameters. The mass flow of wood pellets was kept constant at 100 kg/h using a connecting belt weigher and PTFE was added to the connecting belt at a rate of 0.3 wt% b 300 g/h from a small dosing feeder to ensure uniform blending of the PTFE and wood pellets (Appendix Fig. 1). The range of fluorine concentration in typical municipal waste is 0.010%e0.035% (w/w dry solids) in Germany (EC, 2006). The level of PTFE for the study was chosen to maximize Table 1 PFAS with procedural quantitation limits. Compound Perfluorobutanoic acid Perfluoropentanoic acid Perfluorohexanoic acid Perfluoroheptanoic acid Perfluorooctanoic acid Perfluorononanoic acid Perfluorodecanoic acid Perfluoroundecanoic acid Perfluorododecanoic acid Perfluoro-tridecanoic acid Perfluorotetradecanoic acid Perfluorobutanesulfonic acid Perfluorohexanesulfonic acid Perfluoroheptanesulfonic acid Perfluorooctanesulfonic acid Perfluordecanesulfonic acid Perfluorooctanesulfonamide N-Methyl- Perfluorooctanesulfonamide N-Ethyl- Perfluorooctanesulfonamide N-Methyl-Perfluorooctane- sulfonamidoethanol N-Ethyl-Perfluorooctane- sulfonamidoethanol 1H,1H,2H,2H-Perfluoro- octanesulphonic acid 2H,2H,3H,3H-Perfluoro- undecanoic acid Perfluoro-3-7-dimethyl octane carboxylate 7H-Dodecafluoro heptane carboxylate 2H,2H-Perfluoro decan carboxylate 1H,1H,2H,2H-Perfluorohexan-1-ol 1H,1H,2H,2H-Perflourooctan-1-ol 1H,1H,2H,2H-Perflourodecan-1-ol 1H,1H,2H,2H-Perflourododecan-1-ol Trifluoroacetic acid CAS number 375-22-4 2706-90-3 307-24-4 375-85-9 335-67-1 375-95-1 335-76-2 2058-94-8 307-55-1 72629-94-8 376-06-7 375-73-5 355-46-4 375-92-8 1763-23-1 335-77-3 754-91-6 31506-32-8 4151-50-2 24448-09-7 1691-99-2 27619-97-2 34598-33-9 2043-47-2 647-42-7 678-39-7 865-86-1 76-05-1 Abbreviation PFBA [PFC C4] PFPeA [PFC C5] PFHxA [PFC C6] PFHpA [PFC C7] PFOA [PFC C8] PFNA [PFC C9] PFDA [PFC C10] PFUdA [PFC C11] PFDoA [PFC C12] PFTrDA [PFC C13] PFTeDA [PFC C14] PFBS [PFS C4] PFHxS [PFS C6] PFHpS [PFS C7] PFOS [PFS C8] PFDS [PFS C10] PFOSA N-Me-FOSA N-Et-FOSA N-Me-FOSE alcohol N-Et-FOSE alcohol 1H, 1H, 2H, 2H- PFOS 4HPFUnA PF-3,7-DMOA HPFHpA H2PFDA 4:2 FTOH 6:2 FTOH 8:2 FTOH 10:2 FTOH TFA Quantitation limit mg/m3 6 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 6 0.3 24 24 24 24 0.4 900 K. Aleksandrov et al. / Chemosphere 226 (2019) 898e906 Fig. 1. Schematic of the BRENDA pilot combustion facility at KIT. the mass fraction of PTFE to fuel while staying well below the 1% total halogen limit set by regulations (17 BImSchV, 2003). Kiln incline and rate of revolutions were selected in a way to ensure that heat up, drying, pyrolysis and char burnout of the feedstock could be fulfilled (Gehrmann, 2005). The combustion gases from the rotary kiln were fed into the post combustion chamber equipped with two combined burners for gases and secondary combustion air. The burners (D4.1 and D4.2) were staggered anti-parallel to each other. This configuration allowed for high turbulence and improved mixing of the combustion gases. Regarding the influence of temperature and residence time in the post-combustion chamber on the decomposition rate of PFAS, the basic load of the rotary kiln was kept constant at approx. 1 MW, while the natural gas burners (D4.1 and D4.2) adjusted the temperature and residence time in the post-combustion chamber (Fig. 1). The experiment employed two post combustion chamber conditions to account for partial load (S1) and full load (S2) scenarios common at waste incineration facilities. The low load scenario (S1) ran at a temperature of 870 C and residence time of 4.0 s, while the full load scenario (S2) ran at 1020 C for 2.7 s. To characterize the combustion behavior of the flue gases, samples were extracted via water-cooled lances at the level E0 (after the rotary kiln) and at the level E2 (after the supply of natural gas and air in the post combustion chamber). The hot flue gases left the post-combustion chamber and entered the boiler, where they were cooled to approximately 300 C. The flue gas then entered the pollution control devices which consisted of a spray dryer, a fabric filter, two scrubbers, and a SCR catalyst which met German emission regulation requirements (17 BImSchV, 2003). 2.2. Test materials The PTFE pellets used for incineration were provided by W.L. Gore & Associates GmbH, Putzbrunn; the wood fuel pellets by EC Bioenergie GmbH; and natural gas by Enercity Hanover. Detailed analyses of the combustion materials are presented in the Appendix. 2.3. Flue gas sampling methods The flue gas samples were collected after the heat exchanger (boiler) but upstream of the spray dryer which is the beginning of the pollution control systems (see Fig. 1). This was the optimal location to find the compounds of interest because the flue gas temperatures decreased from 850 to 1000 C to below 300 C which allowed for potential condensation reactions, but excluded any dilution and/or extraction of the compounds from the pollution control systems. Two methods were utilized to collect flue gases. The first method, based on USEPA Method 5 was chosen to collect fluorocarbon compounds of interest (see Table 1) during short-time measurements. The second method, based on VDI 2470 which includes filter units outside of the flue gas duct, was chosen to collect fly ash and HF (VDI 2470, 2011) using long time-measurements. 2.3.1. Test setup for fluorocarbons according to EPA method 5 The first method for flue gas collection utilized the isokinetic sampling train shown in Appendix Fig. 2 (USFR, 2016). Stack sampling procedures consistent with EPA Method 5 for stationary source sampling were followed, except the flue traverse collections points were limited to one axis due to obstructions to the secondary axis. Since the primary system flow measurements (from the modified EPA Method 5 sample train) agreed with the flow calculation within 0.06e4.56% and turbulent mixing occurs prior to the sampling zone (validated by high Re-numbers (>105)), this exception to the method did not impact the results of this study. PFAS were sampled using a modified EPA Method 5 sampling train, utilizing three capture technologies, filtration, impinger K. Aleksandrov et al. / Chemosphere 226 (2019) 898e906 901 sampling and solid adsorbent sampling. Each sampling train was broken down into four samples: quartz fiber filter (1 each), 0.15 M NaOH impinger solutions (450e500 ml), XAD-2 Resin PU foam (22e26 g), and a methanol rinse of the train glassware (150e250 ml). Each of the four samples was sent to a third party laboratory for analysis. 2.3.2. Test setup HF-analytics according to VDI 2470 Out-stack measurements of raw gas and post pollution control system gas at large-scale plants are performed according to VDI 2470 which is similar to the EPA Method 5 and is detailed described in the Appendix (Figs. 3 and 4). The main difference between VDI 2470 and EPA Method 5 is the order of flue gas treatment. Per the VDI guideline, an empty impinger is used to collect the condensate after the filter, while the EPA Method uses the first two impingers to collect the condensate. 2.4. Laboratory analysis methods PFAS analyses were carried out via Liquid ChromatographyMass spectrometry (LC-MS/MS) at laboratories that offered commercially validated methods for the listed compounds (Table 1). SGS Institut Fresenius GmbH- Taunusstein, Germany (SGS) performed trifluoroacetic acid (TFA) sample analysis. All other PFAS analyses were completed at Intertek Consumer Goods GmbH, Frth, Germany (Intertek). Intertek performed liquid chromatography mass spectrometry (LC-MS/MS) blank and spike analyses using 0.5 mg/l standard for the majority of the compounds yielding between 43% and 128% recovery depending on the sampling matrix and compound. The fluorotelomer alcohols (4:2 FTOH, 6:2 FTOH, 8:2 FTOH, 10:2 FTOH) were spiked with 25 mg/l yielding between 56% and 156% recovery depending on sampling matrix and compound. While a few of the compounds had low recoveries at the 0.5 mg/l level, the major of spike results were between 70 and 100% recovery. A list of spike recoveries for each of the sample collection matrixes is available in the supplemental data appendix. VDI 2470 sampling analyses included four parts, fly ash total mass concentration, fly ash burnout (total carbon), the fluoride captured from the vapor stream and weight percent fluoride (F) in the fly ash. All analyses for VDI 2470 were carried out by KIT except for the weight percent fluoride in the fly ash, which was carried out by H.C. Starck using a pyrohydrolic separation of the fluorine with the support of total ionic strength adjustment buffer I solution potentiometrically. This method is briefly described in Pyrohydrolysis in the Determination of Fluoride and Other Halides (Ware et al., 1954). Fly ash concentration in the flue gas was determined by the mass difference of the filters implemented in the long term sampling device in accordance with VDI 2470. The burnout of the fly ash (given as total carbon) was determined via a thermogravimetric analysis according to VDI 2465, part 2 and infrared spectroscopy (IR) detection. Analysis of F from the vapor stream was carried out using ion chromatography (IC), on a Thermo ICS 1000. Total fluorine analysis of the wood pellets was performed by Eurofins Lab in Freiberg, Germany (Eurofins) by the means of a bomb digestion and ion chromatography of the captured condensates. The method is described in detail in DIN EN ISO 16994:2016e12. Eurofins is certified according to Deutsche Akkreditierungsstelle (DAkkS) (D-PL-14081-01-00). 2.5. Thermogravimetric analysis (TGA) To estimate the thermal stability of the material, TGA was performed under nitrogen and air atmospheres with different heating rates. For more details on the methods, see the Appendix. 2.6. Quantitation limits (LOQ) With the exception of TFA, quantitation limits were determined by the quality control procedures of the third party laboratories (i.e. Intertek and H.C. Starck). The procedural quantitation limits were calculated using the third party laboratory quantification limits for each sample type (quartz filter, impinger solution (NaOH), adsorbent media and methanol glassware rinse divided by the sample mass fraction then summed up and divided by the volume of flue gas sampled for each compound analyzed (Appendix, Tables 10 and 11). Procedural quantification limits were calculated for TFA based on field blank samples (DIN 32645, 2008). When the LOQ is analyzed, for further calculation the half of the LOQ is used (Japan MOE, 2001). 2.6.1. Statistics Paired t-testing was utilized to determine if the addition of PTFE created a statistical difference from background levels and to minimize potential interference from external sources. Multiple pairs were analyzed and each pair contained two runs (a blank or control run with the incineration system running at the condition settings with 100 kg/h wood pellet solid fuel and a PTFE spiked run with 300 g/h PTFE pellets added to the wood pellet fuel). A 95% confidence interval was set to determine significance. Thus a pvalue of 0.05 was required to determine if a signal could be distinguished when compared to a control (blank) run. It should be noted that all compounds listed in Table 1 were evaluated separately. 3. Results and discussion This chapter is divided into results from analysis of the supplied materials (PTFE, wood pellets), combustion behavior, fluorine mass balance and the results from the PFAS analysis. 3.1. Analysis of the fuels Wood pellet samples were collected from each shift and were analyzed in duplicate by Eurofins (n 15). The primary elemental composition which included carbon and hydrogen showed a low standard deviation between 0.01 for hydrogen through 0.03 for all other elements. All fluorine values were below the detection limit of 0.001% with exception of one sample collected on February 11th, 2018 (see Fig. 8 in the Appendix). In this case a third sample was collected and analyzed to verify that the single detectable analysis was an outlier. The chemical content and the quality of the PTFE granules were proved with the help of Energy Dispersive X-ray Analysis (EDX) and Fourier-Transform Infrared Spectroscopy (FT-IR). The Fluorine and Carbon concentration determined experimentally by the means of EDX (Fig. 9, Appendix) agreed with the theoretical values derived from the stoichiometric formula of PTFE e [C2F4]n, i.e. 33.33 mol-% C and 66.67 mol-% F. In one of the investigated granules, traces of Al (0.24 mol- %) were detected. In this case, contamination potentially occurred during the sample preparation. In the FT-IR spectra of the granules only the characteristic bands of PTFE were presented (Fig. 10, Appendix). The most intense bands at approx. 1200 cm1 matched the stretching vibrations of CF2 at 1211 cm1 and 1154 cm1 (Fazullin et al., 2015). The band below 650 cm1 showed the rolling vibrations and the planar deformation of CF2 (Fazullin et al., 2015). TGA indicated that the PTFE decomposition process appeared to start around 500 C and was complete around 650 C (Appendix, Tables 3 and 4; Figs. 9 and 10). Estimated half-life times at 800 C 902 K. Aleksandrov et al. / Chemosphere 226 (2019) 898e906 (t1/2 ln 2/k for first order reactions) were well below 0.1 s. Com- plete decomposition could be expected approximately at ten halflife times, equating to less than 1 s residence time at this temperature. 3.2. Combustion behavior The mass flow of wood pellets and natural gas as well as the respective air ratios for combustion were adjusted to obtain enough thermal output, to avoid ash melting and to avoid loss of unburnt pellets into the deslagger (see Fig. 1), which was located after the rotary kiln. The air ratio for the natural gas, as the fuel with the greatest thermal output, was set below one (l 0.7), while the main air in the rotary kiln was set to a superstoichiometric value (l 2.5). Thus, the overall air ratio in the kiln was 1.43 without considering air leakage. This stoichiometry setting reduced the NOx-emissions by about 40% compared to the combustion of nat- ural gas (l 2.0) and gasification of wood pellets (l 0.7, Fig. 13 Appendix) at constant total air ratio in the rotary kiln. Frequent visual observation of the solid movement towards the ash discharger confirmed no loss of solids into the discharger. This indicated an almost complete conversion of the PTFE and the wood pellets in the rotary kiln into the gas phase. From the profile measurements across the diameter of the post combustion chamber E2 (see Figs. 1 and 12 in the Appendix) average CO concentration as an indicator for the gaseous burnout for both settings were determined to values equal and below 1 mg/Nm3 referred to 11 vol.-% of O2 independent of the CO release of the rotary kiln (measured at the level E0). The total carbon results, which were analyzed in one fly ash sample taken at S1 after the boiler supported the favorable burnout with 0.25 wt.-% of remaining carbon (see Table 5 in the Appendix). The concentrations of dust were in the range of 8e11 mg/m3 for setting S1 and about 6e7 mg/m3 for setting S2. Please note, the average ash content of the wood pellets was 0.32 wt.-% when tested independently (see Fig. 5 of the Appendix). These results showed that the combustion was very efficient. 3.3. Fluorine balance To generate the Fluorine balance, the dry flue gas flow after the boiler was needed. The measured wet flue gas flow from BRENDA and the water vapor measurement from the IR technique were used to calculate the fluorine balance. The PTFE feeding rate was 300 g/h which corresponds to a mass flow of 228 g/h of fluorine (F). After combustion, fluorine leaves BRENDA in gaseous form, as HF and in solid form, as F-containing ash. Thus, the total fluorine mass flow (total F-export) leaving the system is the sum of the "gaseous" and the "solid" fluorine mass flows. The difference in the water vapor concentration from combustion calculation (see Table 7 in the Appendix) to the other values relates to the missing measurement of the amount of water evaporated from the deslagger water bath after the rotary kiln. From the wet flue gas flow from the process control system, the water vapor flow was subtracted from the measured water vapor concentration by IR technique. This concentration comprised the water from combustion and the evaporation from the deslagger water bath. Higher water vapor concentration, determined e.g. from gravimetric method during the long-term sampling of fly ash reduces the dry flue gas volume flow and lead to a lower fluoride outlet and decreased recovery rate (for the long-term sampling the values of the recovery rate are between 56 and 78 wt%). Low recovery rates were expected since fluorides are very reactive especially with silicates which are a main component of the refractory in BRENDA. The fluorine content in the fly ash could be neglected compared to the HF. The summed data for fluoride capture can be found in Fig. 2. The small black lines above the columns are the Fig. 2. Fluorine output and recovery rate of fluorine based on water vapor concentrations from long-term sampling. K. Aleksandrov et al. / Chemosphere 226 (2019) 898e906 903 errors according the error propagation. The errors were negligible. Please see the Appendix for a detailed discussion of water vapor measurement including error propagation. 3.4. Per- and polyfluoroalkyl compounds (PFAS) For each run, the concentrations of each substance were calculated by adding their masses found on the filter, in the NaOH, in the MeOH and on the XAD-resin and relating this to the dry gas volume sampled. As an example for the calculations, the measured concentrations of all substances and the respective concentrations at LOQ in each matrix are given in Fig. 3 for the control measurement of paired couple 1 (S1). Only the amount of PFOA in MeOH could be quantified to be about 90 ng/Nm3 (dry). All other substances were below the LOQ and therefore assumed to be LOQ for further calculations. By summarizing the concentrations of each substance in all matrices for each run, paired couples, as well as the settings could be compared (see Fig. 4). Due to the varying sample volumes, the LOQ differed for each run. For S1, paired couple 1 is shown. During the paired run, only minor changes in the concentration of PFOA could be observed compared to the control run. Additionally, the concentrations of PFDA and PFDoA were slightly above LOQ. No other substance could be found. For S2, paired couple 8, no substance was visible above LOQ. Generally speaking, no significant differences can be observed between S1 and S2 with respect to the species detected and their concentrations. With those results, paired t-test were conducted. Paired-t-tests are a statistical method to examine the difference of the mean values of two dependent samples and serves to evaluate a hypothesis. In this study, the difference of the concentrations of the PFAS investigated with and without the feed of PTFE was examined. The hypothesis states that the concentrations of the PFAS are independent of the feed of PTFE to the rotary kiln and thus the dispersion around the mean value can either be positive or negative. As a confidence interval, 95% was chosen. Therefore, if the probability value (p-value), which is often used to interpret t-tests, is > 0.05, the hypothesis is correct, and no statistical difference exists between the concentrations with and without the feed of PTFE. For p < 0.05, the concentrations of the PFAS investigated are dependent on the feed of PTFE and the hypothesis is wrong. The detailed results for PFOA for both experimental settings are shown in Table 2. For both settings, the p-values are greater than 0.05, thus there is no statistical correlation in the difference of the concentration of PFOA whether or not PTFE is fed to the rotary kiln. A summary for all PFAS detected in any matrix, for the experimental settings S1 and S2 is given in Table 3 as ng/Nm3 (dry). Only 11 out of 31 compounds were detected. P-values could only be calculated, if the respective substance could be quantified in at least one matrix per measurement and at least 3 paired runs. Otherwise, no calculations could be performed, the PFAS concerned are labelled with "< LOQ". For all PFAS investigated, p-values were larger than 0.05 for either setting, or the concentrations were too low to be quantified Fig. 3. Pattern of the species for for each matrix for the control run of paired couple 1 (S1). 904 K. Aleksandrov et al. / Chemosphere 226 (2019) 898e906 Fig. 4. Pattern of the species for PTFE spiked and control runs for paired couple 1 (S1) and paired couple 8 (S2). Table 2 Results of t and P-values for PFOA. Setting Paired Couple S1 1 2 3 4 5 S2 6 7 8 9 10 11 Type Control PTFE PTFE Control PTFE Control Control PTFE Control PTFE Control PTFE PTFE Control PTFE Control Control PTFE PTFE Control Control PTFE Concentration [ng/Nm3, dry] 189 194 179 169 302 232 270 354 723 184 258 189 644 157 137 159 2743 143 175 143 413 141 Difference (PTFE-Control) [ng/Nm3, dry] 5 10 70 84 539 70 487 22 2600 32 272 t-value 0.624 0.905 p-value 0.564 0.407 by the third party laboratories. Therefore no statistical correlation in the difference of the concentration of the PFAS whether or not PTFE was fed could be determined. Additionally to the experiments at BRENDA spike and blank experiments with PFBA, PFHxA, PFOA, PFDA, PFDoA, PFTeDA and TFA were performed by the KIT and by the third party laboratories. K. Aleksandrov et al. / Chemosphere 226 (2019) 898e906 905 Table 3 Results of all PFAS measured (ng/m3) and P-values for statistical comparison. Abbrev. Setting S1 (870 C & 4 s) Pair 1 Pair 2 Pair 3 Control 0.3% PTFE Control 0.3% PTFE Control PFHxA [PFC C6] PFHpA [PFC C7] PFOA [PFC C8] PFNA [PFC C9] PFDA [PFC C10] PFUdA [PFC C11] PFDoA [PFC C12] PFTrDA [PFC C13] PFTeDA [PFC C14] PFBS [PFS C4] N-Me-FOSE alcohol < LOQ < LOQ 189a < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 194c < LOQ 128a < LOQ 124c < LOQ 102b < LOQ < LOQ < LOQ < LOQ 169c < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 179c < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 232a < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ Abbrev. Setting S2 (1020 C & 2.7 s) Pair 6 Pair 7 Pair 8 Control 0.3% PTFE Control 0.3% PTFE Control 0.3% PTFE PFHxA [PFC C6] PFHpA [PFC C7] PFOA [PFC C8] PFNA [PFC C9] PFDA [PFC C10] PFUdA [PFC C11] PFDoA [PFC C12] PFTrDA [PFC C13] PFTeDA [PFC C14] PFBS [PFS C4] N-Me-FOSE alcohol 154b < LOQ 258c < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 189c < LOQ 145b < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 136b 135b 644c < LOQ 133b 133b < LOQ 134b < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 137b < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 136b a Only found in MeOH, all other concentrations were assumed as 1/2 LOQ. b Only found on Filter, all other concentrations were assumed as 1/2 LOQ. c Only found in MeOH & on Filter, all other concentrations were assumed as 1/2 LOQ. 0.3% PTFE < LOQ < LOQ 302c < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ Pair 4 Control < LOQ < LOQ 270a < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 0.3% PTFE < LOQ < LOQ 354c < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ Pair 5 Control 163b 153b 723c < LOQ 153b 152b 152b < LOQ 154b < LOQ < LOQ 0.3% PTFE < LOQ 156b 184a < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ Pair 9 Control 138b 138b 2743c 128b 130b 128b 128b < LOQ 131b < LOQ < LOQ 0.3% PTFE < LOQ < LOQ 143b < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 141b < LOQ Pair 10 Control < LOQ < LOQ 143b < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 0.3% PTFE < LOQ < LOQ 175c < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ Pair 11 Control 118b 116b 413c < LOQ 117b 115b 115b < LOQ 115b < LOQ < LOQ 0.3% PTFE < LOQ < LOQ 141b < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ 140b p - value < LOQ < LOQ 0.564 < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ < LOQ p - value 0.368 0.424 0.407 < LOQ 0.536 0.571 < LOQ < LOQ < LOQ < LOQ < LOQ It turns out recovery rates are dependent on the matrix and the carbon chain length. For detailed results, please see the appendix (Table 12). 4. Conclusion Of the 31 PFAS species studied only 11 were detected. When comparing the PFAS measurements, for the few compounds detected, no difference from baseline/control levels using paired ttesting for significance could be distinguished. Based on the PFAS levels detected and the randomness of the occurrence throughout the study, it is likely that the source of these signals are due to contamination of the samples from the environment. With proce- dural quantitation limits between 0.3 and 24 mg/Nm3 depending on compound and volume captured (see Table 1), these results give no significant evidence that the PFAS studied (Table 3) were created during the incineration of PTFE could be found. Therefore, it can be expected that municipal incineration of PTFE using BAT is not a significant source of studied PFAS and should be considered an acceptable form of waste treatment. Acknowledgements The authors would like to thank Lothar Sinn, Werner Baumann, Sonja Mlhopt, Sonja Oberacker, Timo Back, Christian Jnger, Kai Mannweiler, Helmut Reis, Alexander Neumaier, and Marvin Trauth from KIT and Fred Carter and Kenny Raughley of W.L. Gore & Associates for help with experimental work; Bernd Zimmerlin and his team from BRENDA for operating 24 h a day; Siegfried Kreisz, for his support regarding the chemistry of fluorides; Catherine Parmeter, Markus Weiser, Rainer Kasemann, Sebastian Bauer of W.L. Gore & Associates for technical assistance; Jochen Hirschfeld for photography. The authors would also like to thank the many academic and industry experts who provided guidance in the development of the study plan. The authors declare no conflicts of interest. Data Availability e All data and information used in this manuscript have been made available by the authors and are included in the paper and the Supplemental Data (Appendix). Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.chemosphere.2019.03.191. Disclosure statement D. Pigeon is an employee of W.L. Gore, a global manufacturer of products made with fluoropolymers. K. Aleksandrov, H.J. Gehrmann, M. Hauser, H. Matzing, D. Stapf, and M. Wexler are affiliated with Karlsruhe Institute of Technology (KIT). KIT was retained and compensated by W.L. Gore to assist in conducting this study and writing of this paper. W.L. Gore & Associates, Inc. and Karlsruhe Institute of Technology donated employee and student time to support this work. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. References [17. BImSchV 2003] Federal Republic of Germany, 2003. Siebzehnte Verordnung zur Durchfhrung des BundesImmissionsschutzgesetzes (Verordnung ber die Verbrennung und die Mitverbrennung von Abfallen - 17. BImSchV) (Seventeenth [Ordinance on the Implementation of the Federal Emission Control Act (Ordinance on Incineration and Co-Incineration of Waste)]). 906 K. Aleksandrov et al. / Chemosphere 226 (2019) 898e906 Arito, H., Soda, R., 1977. Pyrolysis products of polytetrafluoroethylene and polyfluoroethylenepropylene with reference to inhalation toxicity. Ann. Occup. Hyg. 20 (3), 247e255. DIN32645, 2008, 2008. Chemical Analysis e Decision Limit, Detection Limit and Determination Limit under Repeatability Conditions e Terms, Methods Evaluation. Deutsches Institut fur Normung. European Commission, 2006. Integrated Pollution Prevention and Control Reference Document on the Best Available Techniques for Waste Incineration. August 2006. Council European Commission, 2010. Directive 2010/75/EU of the European Parliament and of the Council of 24 November 2010 on Industrial emissions (integrated pollution prevention and control). Off. J. Eur. Union L 334, 17e119 [cited 2018 October 2]. https://eur-lex.europa.eu/legal-content/EN/TXT/? uricelex%3A32010L0075. Ellis, D., Martin, J., Muir, D., Mabury, S., 2003. The use of 19F NMR and mass spectrometry for the elucidation of novel fluorinated acids and atmospheric fluoroacid precursors evolved in the thermolysis of fluoropolymers. Analyst 128, 756e764. Fazullin, D.D., Mavrin, G.V., Sokolov, M.P., Shaikhiev, I.G., 2015. Infrared spectroscopic studies of the PTFE and nylon membranes modified polyaniline. Mod. Appl. Sci. 9 (1), 242e249. Gehrmann, H.J., 2005. Mathematische Modellierung und experimentelle Untersuchungen zur Pyrolyse von Abfallen in Drehrohrsystemen (Mathematical Modeling and Experimental Investigations of the Pyrolysis of Waste at Rotary Kilns). Bauhaus University Weimar, Weimar, Germany. Gehrmann, H.J., Hiebel, M., Simon, F.G., 2017. Methods for the Evaluation of Waste Treatment Processes [cited 2018 October 2]. https://doi.org/10.1155/2017/ 3567865. Henry, B.J., Carlin, J.P., Hammerschmidt, J.A., Buck, R.C., Buxton, L.W., Fiedler, H., Seed, J., Hernandez, O., 2018. A critical review of the application of polymer of low concern and regulatory criteria to fluoropolymers. Integr. Environ. Assess. Manag. 14, 316e334. [IEA 2014] IEA Bioenergy, 2014. Waste to Energy Summary and Conclusions from the IEA Bioenergy ExCo71 Workshop, 2013 21 05; Cape Town (South Africa). http://www.ieabioenergy.com/wp-content/uploads/2014/03/ExCo71-Waste-toEnergy-Summary-and-Conclusions-28.03.14.pdf. Japan MOE, 2001. Manual on Determination of Dioxins in Ambient Air. Japan Ministry of the Environment. 2001 [cited 2018 October 2] https://www.env.go. jp/en/chemi/dioxins/manual.pdf. Nolte, M., Eberhard, M., Kolb, T., Seifert, H., 2005. Incineration of Drums in a Rotary Kiln eA New Control System for Reduction of CO-Emission; Forschungszentrum Karlsruhe GmbH. Institute for Technical Chemistry, Galveston, TX. Thermal Waste Treatment Division, IT3 Conference, May 9-13, 2005. Prevedouros, K., Cousins, I., Buck, R., Korzeniowski, S., 2006. Sources, fate and transfer of perfluorocarboxylates. Environ. Sci. Technol. 40 (1), 32e44. Rankin, K., Mabury, S.A., Jenkins, T.M., Washington, J.W., 2016. A North American and global survey of perfluoroalkyl substances in surface soils: Distribution patterns and mode of occurrence. Chemosphere 161, 333e341. Sinclair, E., Kim, S., Akinleye, H., Kannan, K., 2007. Quantitation of gas-phase perflouroalkyl surfactants and fluorotelomer alcohols released from nonstick cookware and microwave popcorn bags. Environ. Sci. Technol. 41 (4), 1180e1185. Taniyasu, S., Kannan, K., So, M.K., Gulkowska, A., Sinclair, E., Okazawa, T., Yamashita, N., 2005. Analysis of fluorotelomer alcohols, fluorotelomer acids, and short- and long-chain perfluorinated acids in water and biota. J. Chromatogr. A 1093 (1e2), 89e97. Taylor, P., Yamada, T., Striebich, R., Graham, J., Giraud, R., 2014. Investigation of waste incineration of fluorotelomer-based polymers as a potential source of PFOA in the environment. Chemosphere 110, 17e22. U.S. Environmental Protection Agency, 2000. EPA and 3M Announce Phaseout of PFOS. News Release, Washington (DC) [cited 2018 October 2]. https://archive. epa.gov/epapages/newsroom_archive/newsreleases/ 33aa946e6cb11f35852568e1005246b4.html. [USFR] US Federal Register, 2016. Standards of Performance for New Stationary Sources e EPA Method 5 e Determination of Particulate Matter Emissions from Stationary Sources. Appendix A-3 to Part 60 e Test Methods 4 through 5I. Final rule. 40 FR Part 60 Appendix A-3. 2016 Aug 30. 81 FR 59800. Van Belle, G., Fisher, L., Heagerty, P., Lumley, T., 1993. Biostatistics: A Methodology for the Health Sciences, second ed. John Wiley & Sons, Inc., Hoboken (NJ). [VDI 2470] Verein Deutscher Ingenieure, 2011. Messung Gasformiger Emissionen; Messen Gasformiger Fluor-Verbindungen; Absorptions-Verfahren (Measurement of gaseous emissions; Measuring gaseous fluorine compounds; Absorption method). VDI 2470. Ware, J., Cline, W., Tevebaugh, R., 1954. Pyrohydrolysis in the determination of fluoride and other halides. Anal. Chem. 26 (2), 342e346.